AI Chip Giants Compete for TSMC's 3nm Technology

Intel, NVIDIA, and AMD are set to adopt TSMC's 3nm technology in their next-generation accelerators, paving the way for large-scale applications in the artificial intelligence (AI) sector. This advanced 3nm node has become highly sought after, largely due to mainstream tech companies developing their upcoming product lines around it. Notable examples include Apple's A19 Pro chip, MediaTek's Dimensity 9400, and Google's Tensor G5.

The involvement of AI technology giants like NVIDIA and AMD in TSMC's 3nm process is becoming increasingly clear. A new report highlights the significant role this technology will play in their AI product roadmaps. TSMC's 3nm node is poised to be one of the company's most successful processes, driven by high demand in the AI market.

AMD plans to leverage TSMC's 3nm technology for its next-generation Instinct MI355X AI accelerator, which will be based on the upcoming CDNA 4 architecture, promising exceptional performance. Meanwhile, NVIDIA is expected to integrate the 3nm process into its upcoming "Rubin" architecture, set to launch in 2026. NVIDIA is reportedly taking a more measured approach, choosing not to rush the release of its Rubin AI products to compete with other industry alternatives. However, rumors suggest that an AI PC system-on-chip (SoC), jointly developed by NVIDIA and MediaTek and utilizing the 3nm node, could arrive by 2025.

On the other hand, Intel's Falcon Shores architecture, eagerly anticipated in the AI space, may face some challenges. Intel is expected to move away from its own foundry services in favor of TSMC's 3nm technology, positioning its AI products more competitively. Falcon Shores is critical to Intel's AI business and long-term strategy.

With growing adoption in mobile, automotive, and AI industries, TSMC's 3nm process is set to generate substantial revenue, potentially becoming one of the company's most successful products to date.